Numerous miniaturized sensors and actuators have been fabricated since the emergence of MEMS technology. Since the majorities of MEMS fabrication processes are either directly borrowed or derived from IC technology. Inherently, most MEMS devices are built on rigid substrates such as silicon and glass wafers. On the other hand, for a wide variety of applications, it has long been desirable that sensors, actuators, or circuits can be fabricated on flexible substrates so as to be mounted on nonplanar surfaces or even on flexible objects such as a human body. One example is the tactile sensor, which need to be flexible in order to be attached to curved shapes like fingers and arms. In the field of fluid monitoring and controlling, it is often of great interest to know the profile of certain physical parameters such as pressure, or shear stress distributions on a nonplanar surface. To address the challenges arising from these applications, we need to develop a technology which can enable the fabrication of micromachined sensors on flexible substrate, namely, a MEMS skin technology. Additionally, integration of circuits is highly desirable since this integration promises to bring very important benefits such as operational improvement, packaging simplification, and cost reduction. Therefore, we need to develop a flexible skin technology which is compatible with both ICs and MEMS.
There are many methods to fabricate flexible miniaturized transducers. The most straightforward method to make flexible transducers/electronics is to fabricate directly on flexible substrates similar to the fabrication of thin film transistors on plastic/polymer/metal substrates. The advantages of this method are low cost and the ability to make large area flexible structures. Nevertheless, due to the temperature limit imposed by the flexible substrates, many high temperature processes are ruled out and the material properties are not optimized. Transducers that need high temperature process or use rigid materials such as single crystal silicon are difficult to fabricate on flexible substrate. Furthermore, electronics can not be integrated using the mainstream IC technology. Although circuits based on amorphous silicon/conductive polymers are active research topics now, they will not catch up with the complexity and performance of circuits based on single crystalline silicon in the foreseeable future. Moreover, when the substrate is subjected to bending, the devices on it undergo the stress as well. This may cause two undesirable consequences: 1) the devices on the flexible substrates may crack if the bending curvature is too large; 2) the performance of the devices is affected by the substrate bending.
Dr. Xu developed a unique silicon flexible skin technology that is totally different from the traditional method. The conceptual fabrication process is illustrated in the following figure. Assuming that MEMS devices or ICs have already been fabricated on the silicon substrate, the first step of the skin fabrication is to coat a polymer layer on the front of the wafer. Then the polymer layer is patterned to expose metal pads. Note that if necessary, MEMS and ICs can be exposed as well at this step. After this, the silicon wafer is thinned down and etched through from the back to form the arrays of silicon islands by Deep Reactive Ion Etching (DRIE). Finally, another layer of polymer is coated on the back to encapsulate the silicon islands.
The basic structure of the silicon flexible skin is arrays of silicon islands sandwiched between two layers of polymers. MEMS devices and ICs are on rigid islands. When the skin is bent, the devices on islands will not be subjected to stress. At the same time, the islands are small enough that the flexibility is not impaired. The most important advantage of this technology is its compatibility with current MEMS and IC technologies, since MEMS devices and ICs can be fabricated on the silicon wafer before the formation of the skin. Not only significant R&D efforts can be saved by avoiding re-invention, but also abundant sensing and computation capabilities offered by the silicon-based technology can be readily integrated.
Using this technology, Dr. Xu developed flexible shear-stress sensor skins for flow separation detection. The sensor skin contains a 1-D array of 36 shear-stress sensors, which can cover the 180¡ã surface of the half-inch diameter semi-cylinder with 5¡ã resolution (Fig. 1). These sensor skins have been installed on an Unmanned Aerial Vehicle (UAV) and flight-tested successfully (Fig. 2). Real-time flow distribution around the leading edge was obtained by the sensor skin.
Figure 1. The shear-stress sensor skin mounted on a 0.5¡± diameter aluminum block. Reprinted from .
Figure 2. The unmanned aerial vehicle installed with the shear-stress sensor skin. Reprinted from 
The first IC-integrated flexible shear-stress sensor skin
has also been demonstrated using a post-
Figure 3. One IC-integrated skin held in tweezers. Reprinted from .
aforementioned work was carried out by Dr. Xu at the Micromachining Lab at the California
Institute of Technology. At
Figure 4. Simplified process flow: (a) Boron diffusion; (b) Patterning the device layer and removing the exposed BOX layer; (c) Al deposition and patterning to form traces and pads; (d) 1st 3 mm parylene deposition; (e) Patterning the parylene openings and etching away underneath metal traces; (f) XeF2 etching to release the devices; (g) 2nd 10 mm parylene deposition; (h) Patterning the parylene layer and releasing the device. Reprinted from 
Fig. 5 (a) shows bent flexible sensor skin fabricated using the new technology. Four MOSFETs with different channel widths integrated on the flexible skin can be clearly observed in Fig. 5(b). The 2-D arrays of marks are the re-sealed etching holes in parylene film. Fig. 5 (c) is a SEM image of a MOSFET embedded in the parylene flexible substrate. The functionality of the flexible MOSFET has been successfully demonstrated, and its sensitivity to bending has also been characterized.
Figure 5. (a) A bent flexible device held by a pair of tweezers; the inset is an optical micrograph of four MOSFETs with different channel widths. (b) SEM image of a MOSFET integrated on the flexible substrate. Reprinted from 
Note that during step (g), two physically separated parylene C layers are formed. One is deposited on the bottom of the XeF2-etched silicon surface, and the other layer encapsulates the devices. In this case, the bottom parylene layer was removed to fabricate a flexible skin with single crystal silicon MOSFETs. For other applications, we can choose to keep the bottom parylene C layer. Therefore, microchannels, microtubes or diaphragms can be readily integrated. We have demonstrated smart tubes and smart yarns using this approach. Fig. 6(a) shows an optical image of a fabricated smart tube device. The whole device is made out of conformally coated parylene C which makes the device highly transparent. The pressure sensor is located on the square diaphragm on the left, and is composed of four silicon piezoresistors placed in direction perpendicular to the edge to maximize the strain during the diaphragm deformation. The flow sensor, which is based on thermal principles and made of a silicon heater, is placed further to the right of the tube. The 2D arrays of white dots are the sealed etching holes on parylene. The backside SEM image of the device is illustrated in Fig. 6 (b). The scallop-shape is the result of isotropic etching of XeF2. Fig. 6 shows a strand of knotted smart yarn fabricated based on a long parylene tube. To strengthen the mechanical property and make it kink-free, the parylene tube is filled with PDMS as shown in the inset.
Figure 6. (a) Optical image of the fabricated smart tube device including both pressure sensor and flow sensor; (b) back side and cross-section SEM images of the device. Reprinted from .
We believe that flexible skin technologies will play crucial roles in the development of advanced medical implants, wearable sensors, and intelligent textiles. With post-CMOS and post-MEMS compatible technologies, it is expected that multi-functional flexible systems with high-performance sensors and electronics will be developed. More detailed information of post-CMOS and post-MEMS compatible smart skin technologies can be found in Dr. Xu¡¯s review paper which will appear on the special issue on Flexible Sensors and Sensing Systems for the IEEE Sensors Journal .
The work carried out at Wayne State University was supported by the National Science Foundation under Grant No. 0747620. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.
8. Y. Xu, ¡°Post-CMOS and Post-MEMS Compatible Flexible Skin Technologies: A Review¡±, IEEE Sensors Journal (in press)